Skip to content

v1.0.0

Choose a tag to compare

View all tags
@FlorentB974 FlorentB974 released this 26 Sep 10:17
· 39 commits to main since this release
v1.0.0
2f352f8

🎉 graphrag v1.0.0 — First public release

Date: 2025-09-26

Welcome to the very first official release of graphrag! This release marks a milestone — the core Retrieval-Augmented Generation (RAG) pipeline is ready for use. Thank you to everyone who helped test, file issues, and contribute ideas. Let's jump into what's included.

Highlights

  • Core RAG pipeline: ingest documents, chunk them, compute embeddings, store them in the graph database (Neo4j), retrieve relevant chunks, and generate grounded answers using OpenAI or Ollama LLM backends.
  • Multi-format ingestion: the loaders/ and ingestion/ modules support PDFs, DOCX, PPTX, CSV and plain text files, with a pluggable loader system.
  • Graph-backed retrieval: document chunks are stored in a Neo4j graph to enable relationship-aware retrieval and visualization of document relationships (see core/graph_db.py and core/graph_viz.py).
  • LLM integration: a unified LLMManager (core/llm.py) supports OpenAI and Ollama providers with clean fallbacks and response post-processing to remove HTML/LaTeX artifacts.
  • Chainlit and web UI wiring (starter): interactive UI wiring is included to let you upload documents and ask questions in real time.
  • Dev-friendly scripts: scripts/ingest_documents.py, scripts/create_similarities.py, and scripts/setup_neo4j.py help bootstrap datasets and indexes quickly.

Fixed reported issues

  • Improved HTML and LaTeX cleanup in generated answers (see core/llm.py) to ensure plain-text responses without broken markup.
  • Cleaner chunking heuristics and improved default chunk size/tokens in core/chunking.py.
  • Better error handling when LLM calls fail, with logged errors and clearer exceptions.
  • Docker/CI: initial workflow to build and publish images to GHCR was added and documented.

Files of interest

  • core/llm.py — LLMManager with generate_rag_response and cleaning helper
  • core/chunking.py — chunking utilities and heuristics
  • core/graph_db.py & core/graph_viz.py — Neo4j integration and visualization helpers
  • ingestion/document_processor.py & loaders/* — document ingestion pipeline
  • scripts/ — helper scripts for ingestion, similarity creation and Neo4j setup

Upgrade notes / breaking changes

  • This is the first release; there are no prior versions to migrate from. Default settings are in config/settings.py — double-check OpenAI/Ollama API settings and Neo4j connection info before ingesting large datasets.

Thank you

This project exists because of many helpful testers and idea contributors. If you find issues, please open them on GitHub — your feedback helps shape the roadmap.

Happy building,

The graphrag contributors

Assets 2
Loading